Graphical methods for evaluating covering arrays

Abstract

Covering arrays relax the condition of orthogonal arrays by only requiring that all combination of levels be covered but not requiring that the appearance of all combination of levels be balanced. This allows for a much larger number of factors to be simultaneously considered but at the cost of poorer estimation of the factor effects. To better understand patterns between sets of columns and evaluate the degree of coverage to compare and select between alternative arrays, we suggest several new graphical methods that show some of the patterns of coverage for different designs. These graphical methods for evaluating covering arrays are illustrated with some examples. Copyright © 2015 John Wiley & Sons, Ltd.

Selecting the Best Wild Card Entries in a Covering Array

“…This will allow the experimenter to choose the best combination of wild card entries to match their priorities. Once a best covering array is chosen, then the three‐way coverage missing region plot and correlation‐based plot suggested by Kim et al can help summarize the pattern of coverage of the covering array.…”

“…In Section 2, we briefly describe the basic criteria that we use for evaluating the covering arrays, including those proposed by Kim et al In Section 3, we illustrate the method with some examples to find the best CA designs subject to the selected criteria. Section 4 contains some discussion and conclusions.…”